Organic solar cells are less efficient than their silicon counterparts, but ccould be cheaper and more versatile. Credit: T. Wang/

US researchers have used computer modelling to identify an organic molecule with useful electrical properties - proof-of-concept for an approach that could soon yield new compounds to harvest solar energy in photovoltaic cells.

Alán Aspuru-Guzik, a theoretical chemist at Harvard University in Cambridge, Massachusetts, and his colleagues, used computational models to screen a family of organic molecules and identify those likely to be the best semiconductors. The team passed the finding to researchers at Stanford University in California, who have now synthesized the molecule and confirmed its properties.

The results are encouraging for Aspuru-Guzik, who, in collaboration with computer giant IBM, is using the same computational tools to screen some 3.5 million organic molecules in the search for a new generation of flexible and lightweight solar cells. His team plans to publish the structures of the 1,000 molecules with the most useful calculated properties, in an effort to help bench chemists focus on the best structures to synthesize.

The project, which has been running for more than two years, employs some of the same methods used by drug companies. "It's how the pharmaceutical people do it: the theorists give a ranking to the experimentalists," says Aspuru-Guzik. "We're trying to save experimental time."

From theory to reality

The latest molecule is one of the best organic semiconductors yet discovered, in terms of its ability to transport electric charge. The study is published in Nature Communications today1.

"It's beautiful work," says Thuc-Quyen Nguyen, a chemist at the University of California, Santa Barbara. Chemists have typically focused on individual compounds or families but systematic screening could save time and reveal new opportunities, she says. "That's the novelty of the approach."

Aspuru-Guzik started the screening process by identifying a known organic semiconductor with desirable properties. He then designed some possible derivatives of that compound, and used quantum and molecular mechanical models to predict their properties. His team passed the structure of the best candidate along to Zhenan Bao, a synthetic chemist at Stanford, and her colleagues, who spent six months making the chemical and then tested it in an experimental transistor.

Bao's team was pleasantly surprised to discover that the molecule conducted electric charge between three and four times better than predicted. This shows how difficult it is to predict a molecule's properties accurately, but confirms that models are good at ranking molecules according to their relative performance, says Bao.

As experimentalists synthesize theoretically designed molecules in future, their data will help theorists to improve their own models. "This will be an iterative process," she says. "Hopefully the theory will get better and the prediction will be right on target in the future."

Solar screening

In his search for solar cells, dubbed the Clean Energy Project, Aspuru-Guzik is screening molecules for a host of properties involved in converting sunlight into electrical energy. The goal is to provide the materials that will allow organic photovoltaic cells to turn more than 10% of the solar energy that hits them into electricity, compared with about 9% for the best materials today.

That goal is still lower than the conversion ratio achieved by modern cells made from silicon, but organic photovoltaics would be cheaper and could be used in fabrics, plastics and even inks and paints. Commercial organic photovoltaic cells could hit the market within a few years. Some believe their cheapness and versatility will make them especially useful in the developing world.

The Harvard team is running its quantum-mechanical computations through IBM's World Community Grid, which harnesses idle time on volunteers' personal computers. So far, the initiative has inspected 2.3 million molecules, and should reach the 3.5-million mark next year.

That is probably the largest set of quantum calculations ever conducted in chemistry, says Aspuru-Guzik. His team is also running its own calculations, looking at simple structural and chemical properties.

The screening has already begun to reveal some potentially interesting new compounds and structural families, including various molecules containing selenium. The team plans to publish the top 1,000 candidates in the next few months.

Aspuru-Guzik is confident that the list will bear fruit. "If I'm wrong," he adds, "here is my head on the chopping block."